The plant vacuolar sorting receptor (VSR) binds proteins carrying vacuolar sorting signals (VSS) of the ‘sequence-specific’ type (ssVSS) but not the C-terminal, hydrophobic sorting signals (ctVSS). Seeds of Arabidopsis mutants lacking the major VSR isoform, AtVSR1, secrete a proportion of the proteins destined to storage vacuoles. The sorting signals for these proteins are not well defined, but they do not seem to be of the ssVSS type. Here, we tested whether absence of VSR1 in seeds leads to secretion of reporter proteins carrying ssVSS but not ctVSS. Our results show that reporters carrying either ssVSS or ctVSS are equally secreted in the absence of VSR1. We discuss our findings in relation to the current model for vacuolar sorting.
Vacuoles are defining organelles of plant cells. They are the intracellular end-point of the plant secretory and endocytic pathways and can have a multiplicity of functions (1,2). In some plant cells, acidic lytic vacuoles (LV) with degradative functions can coexist with more neutral protein storage vacuoles (PSV), e.g. during developmental transitions in seeds (3,4) or even in the same mature cell (5,6). The existence of distinct vacuolar compartments implies that separate sorting pathways must exist to guarantee protein targeting to the correct vacuole (2,7). Indeed, different types of vacuolar sorting signals (VSS) have been identified and assigned to a specific sorting route (8). These signals have been grouped into sequence-specific (ssVSS), C-terminal (ctVSS) and physical structure signals (9). ssVSS are contained in a propeptide, located anywhere in the proprotein, and include a conserved motif [N/L]-[P/I/L]-[I/P]-[R/N/S] (10), which contains a crucial aliphatic residue (isoleucine or leucine). Mutation of this residue results in protein secretion (11).
ssVSS are ligands of a family of vacuolar sorting receptors [VSRs; (12,13)]. Vacuolar sorting receptors shuttle between the trans Golgi network (TGN) where they bind ssVSS-bearing cargo and the prevacuolar compartment (14,15) where cargo is released. Vacuolar sorting receptors can recruit clathrin adaptors (15,16). Therefore, they are thought to sequester LV-bound proteins into clathrin-coated vesicles. ctVSS, on the other hand, have no conserved motif but consist of a short string of hydrophobic residues located at the very C-terminus of the protein (8,17). They do not bind VSR and are suspected to trigger homotypic aggregation of cargo (18). An as-yet uncharacterized interaction with the Golgi membrane, possibly involving a different kind of receptor – the receptor homology-transmembrane-RING H2 domain protein family (19,20)– results in sequestration into ‘dense’ vesicles (DV) and eventual targeting to the PSV (21) in a mechanism akin to sorting by aggregation in secretory granules (22).
The conceptual separation of VSR- and DV-based sorting routes pleased everyone until exceptions began to appear. For example, some bona fide storage proteins were found to contain ssVSS (23,24) that bound to VSR in vitro (25), suggesting that VSR could also be involved in the sorting of storage proteins to PSV. Strong support for this hypothesis came from the evidence that genetic knockout of the major seed isoform of VSR in Arabidopsis (AtVSR1) led to the missorting of endogenous vacuolar storage proteins into the apoplast (26). More recently, a green fluorescent protein (GFP) reporter protein carrying a ‘compound’ VSS of β-conglycinin (27,28) was also shown to be missorted and used as the basis for an elegant genetic screen for PSV-trafficking mutants (29).
A central role for VSR in sorting proteins to seed PSV would be easier to rationalize if seed cells only contained one type of vacuole, the PSV. A single vacuolar compartment would then be the obvious common destination for multiple sorting pathways, including the VSR-mediated one. Indeed, a number of groups have now convincingly established that during the period of storage protein deposition, Arabidopsis seeds do indeed contain a single type of vacuole, the PSV (19,30,31).
From the above, it appears undeniable that VSR1 is important for PSV sorting in Arabidopsis seeds, as further demonstrated by the fact that the mutant screen based on the secretion of PSV-directed GFP repeatedly identified several mutant alleles of VSR1 (29). However, there are problems with this model. First, the segregation of storage proteins into DV is initiated in the cis Golgi (19,32,33), whereas VSR is concentrated at the trans Golgi (19). Second, the absence of VSR1 results in the missorting of only a proportion but by no means all storage proteins (26). Third, it is not clear that VSR can bind a ‘canonical’ ctVSS, which seems to be the norm on PSV-bound proteins. In the case of β-conglycinin, transposition experiments using GFP as a reporter have revealed the presence of both a ctVSS, which is only capable of direction to the vacuole when exposed at the C-terminus of GFP, and a ssVSS (when placed at the N-terminus of GFP), which is only functional when the core Ser–Ile–Leu residues are intact. Therefore, it is difficult to evaluate the individual contributions of these sequences to vacuolar sorting (27,28).
The VSS of endogenous Arabidopsis storage proteins have not yet been defined, although a C-terminal peptide from 12S cruciferin was shown to bind VSR in vitro (26). We hypothesize that if VSR1 is the true receptor for ssVSS-bearing proteins (which might include all Arabidopsis storage proteins), its downregulation should lead to the missorting of this class of proteins only, but not of proteins bearing a C-terminal type of sorting signal. Here, we have tested this hypothesis. We have studied the effect of the absence of VSR1 on vacuolar transport of reporter proteins bearing unambiguous ssVSS or ctVSS in Arabidopsis seeds. Our results indicate that the absence of VSR1 results in missorting of both classes of proteins. This suggests that VSR function may not be to directly bind and sort storage proteins and that VSR1 absence instead leads to pleiotropic trafficking defects.
Results and Discussion
We initially used a set of reporter proteins (Figure 1A) based on the bean storage protein phaseolin, mutated to contain only one of its two glycosylation sites (T343F: 34). This simplifies the electrophoretic mobility pattern of the protein without affecting its intracellular fate (34). T343F is successfully transported to the vacuole in tobacco and Arabidopsis leaf protoplasts and transgenic plants, as indicated by the appearance of diagnostic proteolytic fragments (34,35). We generated transgenic Arabidopsis plants expressing T343F and subjected seed homogenates to immunoblotting with a phaseolin antiserum. We detected a polypeptide of the size expected for intact phaseolin as well as the diagnostic proteolytic fragments (Figure 1B, arrowheads and vertical bar, respectively). This indicated that the majority of T343F was delivered to PSV and was further confirmed by immunoelectron microscopy (Figure 4E). T343F appears to partition into regions of the PSV matrix, which are more electron-opaque than their surroundings (Figure 4E). This is likely because of phaseolin’s propensity to form aggregates during transit and therefore arrive at the vacuole in aggregate form (18,36), a feature shared by other storage proteins (37).
T343F is sorted to the vacuole by virtue of its C-terminal tetrapeptide AFVY, which constitutes the vacuolar sorting motif (36,38). Deletion of this signal to generate Δ418 (Figure 1A) results in the protein being secreted by tobacco protoplasts (36). In Arabidopsis seeds, Δ418 is detected only as the intact precursor (Figure 1B), indicating that this protein has not reached the vacuole. When a well-studied ssVSS, the 12-residue ‘linker’ peptide from proricin (24), is appended to Δ418 (to give Δ418-L; Figure 1A), vacuolar targeting is restored (39). Likewise, in Arabidopsis seeds, Δ418-L undergoes proteolytic processing in vacuoles (Figure 1B, vertical bar).
These data show that two phaseolin-based reporters, each carrying different types of VSS, have the same fate (i.e. vacuolar delivery) in Arabidopsis seeds. We have recently shown that both AFVY and the ricin linker can target the monomeric red fluorescent protein (mRFP) to PSV in seeds (30), thus validating the efficacy of both sorting signals in this heterologous system.
Having established that two different types of sorting signals can target phaseolin to the vacuole, we set out to test whether vacuolar delivery of these reporters would be affected in seeds from mutant Arabidopsis (vsr1-1) lacking the major seed-specific isoform of the VSR, VSR1 (26). We therefore crossed transgenic plants expressing phaseolin- or red fluorescent protein (RFP)-based reporters with vsr1-1 knockout plants and selected F2 plants homozygous for vsr1-1. We then studied the processing and intracellular distribution of the reporters (Figure 2). In vsr1-1 seeds, both T343F and Δ418-L produced a lesser amount of proteolytic fragments, indicating that their vacuolar delivery may be inhibited (Figure 2A, compare lanes 4–6 with lane 3 and lanes 10–12 with lane 9). The reduction in vacuolar processing was paralleled by an increase in the intensity of their precursor polypeptides (Figure 2A, arrowheads).
As a control for the lack of VSR1 in the mutant background, we separated proteins on SDS–PAGE and stained them with Coomassie Blue (Figure 2B). In vsr1-1, native Arabidopsis storage proteins 2S albumin and 12S cruciferin are seen to be partially missorted, resulting in the lack of vacuolar processing and the accumulation of their precursor polypeptides (26,29). Figure 2B shows that in the crosses, 12S precursors accumulated, exactly as observed in the parental vsr1-1 line (26).
As the accumulation of precursor polypeptides alone is not sufficient to establish whether the reporters are missorted or simply not processed (i.e. because the processing enzymes themselves are missorted), we analyzed seeds from our reporter lines by immunoelectron microscopy (Figure 3). In seeds of wild-type background, the phaseolin antibody gave very low background in the apoplast (Figure 3G). However, in vsr1-1 background, a significant increase in the number of gold particles per square micrometre in the apoplast was observed (Figure 3, compare panel D with panel C and panel F with panel E), indicating that a proportion of both T343F and Δ418-L are indeed secreted (Figure 3G). This shows that the absence of VSR1 leads to partial secretion of phaseolin-based reporters, regardless of the type of sorting signal they carry. Secretion is indeed partial because in vsr1-1 background, the PSV still contains considerable amounts of either Δ418-L or T343F (Figure 4, panels D and F).
In a complementary approach, we used stable transgenic lines expressing the monomeric red fluorescent protein (mRFP), carrying either the ricin linker ssVSS (L-RFP) or the phaseolin ctVSS AFVY (RFP–AFVY). Both are correctly targeted to PSV in seeds, although L-RFP is secreted when its crucial isoleucine is mutated (30). To study the fate of these reporters in the absence of VSR1, we crossed RFP reporter lines with vsr1-1 lines and imaged embryos dissected from mature seeds of transgenic plants by confocal microscopy. Figure 5 (panels A–D) shows, as reported previously (30), that in the wild-type background, RFP carrying either AFVY or the ricin linker localizes completely to the lumen of the PSV. In the vsr1-1 background, a significant proportion of the RFP reporter, regardless of the type of VSS, is detected in the intercellular spaces (Figure 5, panels E–H). This confirms that the absence of VSR1 leads to missorting of vacuolar proteins regardless of the type of sorting signal they carry.
The indiscriminate secretion of PSV-destined proteins and the fact that a major proportion of this transport cargo still reaches the PSV seem to point to an indirect role for VSR1 in vacuolar protein sorting. VSR1, by virtue of its position at the trans Golgi, may play a salvage role after the main sorting event has taken place (19). Its absence would therefore lead to the secretion of stray cargo that has failed to enter DV. Alternatively, absence of the receptor could affect membrane recycling between the prevacuolar compartment and the Golgi in the long term, resulting in a pleiotropic missorting phenotype. However, if the effect on membrane trafficking is so severe, we would expect targeting of proteins to the tonoplast to be equally affected. To test this, we crossed plants constitutively expressing Arabidopsis α-tonoplast intrinsic protein-yellow fluorescent protein (α-TIP-YFP), an aquaporin that localizes to the tonoplast of seed PSV in Arabidopsis (30,40), with vsr1-1 plants. We compared the distribution of α-TIP-YFP in wild-type or homozygous vsr1-1 background by confocal microscopy. We used PSV autofluorescence (false-coloured blue) to highlight these organelles, as previously described (26,30). Figure 5 (panels L–N) shows that vsr1-1 seeds contain a slightly higher number of smaller PSV. However, α-TIP-YFP is still correctly targeted to the PSV tonoplast, without any evidence of missorting to a different membrane location. Therefore, the lack of VSR1 seems to affect the sorting of soluble vacuolar proteins, but does not lead to major defects in tonoplast protein targeting.
We can rationalize our results by formulating two hypotheses: (i) VSR1 is the true receptor for storage proteins and can bind both ssVSS and ctVSS. In its absence, VSR3 and VSR4, the remaining seed-expressed isoforms, take over. This hypothesis is testable by knocking out these additional isoforms in seeds. However, evidence for the binding of VSR to well-studied ctVSS is still lacking. In particular, no binding of VSR to AFVY has ever been observed (our unpublished observations). Moreover, real-time polymerase chain reaction (PCR) analysis revealed that the absence of VSR1 does not result in transcriptional upregulation of VSR3 and VSR4 (Figure 6). (ii) VSR1 acts as a downstream salvage receptor, and its absence affects recycling of sorting effectors from the PVC, ultimately leading to an imbalance in trafficking events. The missorting of storage proteins would then be a consequence of this pleiotropic alteration in membrane homoeostasis. Indeed, lack of a plant retromer component, VPS29, thought to participate in the recycling of VSR to the trans Golgi (45) causes a phenotype that is almost indistinguishable to vsr1-1 (41).
In our opinion, the latter hypothesis seems more likely. Because the majority of storage proteins (and of our reporters) are still correctly sorted to the PSV, this suggests that the main sorting event for these proteins occurs upstream of the TGN where VSR1 is principally localized.
Materials and Methods
A schematic representation of the constructs used in this study is shown in Figure 1A. All sequences were amplified by PCR and initially cloned into the 35S-CaMV promoter/terminator cassette (46). The complementary DNAs (cDNAs) of T343F, Δ418-L and Δ418 were amplified from existing constructs in pDHA (36,39) using the following forward primer 5′ CCGGATCCATGATGAGAGCAAGGGTTCCA 3′ (BamHI site underlined) in conjunction with the appropriate reverse primers 5′ CCCCCGGGTCAGTACACAAATGCACCCTTTC 3′ (T343F), 5′ CCCCCGGGTCAATTAAAATTTGGTACCACTGGCCTTATAAGCAAAGAGGCACCCTTTCTTCCC 3′ (Δ418-L) and 5′ CCCCCGGGTCAACCCTTTCTTCCCTTTTGCTGTTCCTG 3′ (Δ418). Underlined bases in reverse primers indicate SmaI sites.
The 35S cassette was then digested using EcoRV and subcloned into the binary vector. The construction of spRFP, spL-RFP and α-TIP-YFP has been described previously (30). The binary constructs were transformed into Agrobacterium tumefaciens EHA105 (47).
Plant transformation and crossing
Arabidopsis thaliana Columbia ecotype between growth stages 6.20 to 6.50 (48) was transformed as described (42). Transgenic plants expressing different reporters were crossed with the vsr1-1 mutant (26) between stages 6.10 to 6.40 (48). The F2 progenies were genotyped by PCR to identify homozygous lines using previously described primers for T-DNA and VSR1 (26).
Total seed protein extraction and protein gel blot analysis
For analysis of total proteins, 20 seeds were milled for 10 min in a 1.5-mL tube containing a ball bearing and 200 μL Laemmli loading dye (2×) containing 0.5 m Tris–HCl (pH 6.8), 4.4% (w/v) sodium dodecyl sulphate (SDS), 20% (v/v) glycerol, 0.036% bromophenol blue in distilled/deionized water and 2% (v/v) β-mercaptoethanol. The ground samples were boiled for 5 min, followed by centrifugation for 2 min at 15 000 × g and the supernatant was separated by SDS–PAGE. Immunodetection was performed with anti-phaseolin antiseurm using the enhanced chemiluminescence system (Amersham).
Leaf segments or embryos dissected from mature seeds were mounted in water and imaged with a Leica TCS SP2 confocal laser scanning microscope using a 10× (NA 0.3) air or a 63× (NA 1.4) oil immersion objective. Yellow fluorescent protein was excited at 514 nm and detected in the 525–583 nm range. Red fluorescent protein was excited at 543 nm and detected in the 553–638 nm range. When imaging mature embryos, PSV autofluorescence was excited at 405 nm and detected in the 450–510 nm range (30). Simultaneous detection of PSV autofluorescence and YFP was performed by combining the settings indicated above in the ‘sequential scanning’ facility of the microscope. Images were resized with adobe photoshop CS. No image enhancement was performed.
Fixation of seeds with high-pressure freezing was performed as described previously (19). Dry seeds were imbibed in water for 1 day. The seed coat was removed, and the embryo was submerged in hexadecene. The embryos were mounted in planchettes and frozen in a high-pressure freezer (HPF010; Bal-Tec). Freeze substitution was performed with a Leica AFS freeze substitution unit (Leica) with dry acetone supplemented in 0.1% uranyl acetate at −85°C. The samples were embedded in Lowicryl HM20 resin. Ultrathin sections were cut with a Leica Ultracut S (Leica) and incubated with antibodies against phaseolin at a dilution of 1:2000, followed by an incubation with 10-nm gold-coupled secondary antibodies (BioCell GAR10) at a dilution of 1:50.
Fixation of dry seeds with London Resin White was performed as described previously (43). After removal of the seed coat, the embryos were fixed in 0.25% glutaraldehyde and 1.5% paraformaldehyde, followed by a second fixation in 0.4% osmium tetroxide, and a staining with 1% uranyl acetate. The samples were dehydrated in ethanol and embedded in London Resin White. Ultrathin sections were incubated with the antibody against phaseolin at a dilution of 1:400.
Gold particles in the apoplast were counted in 10 independent sections for each transgenic line. The area of each section was calculated using imagej (rsb.info.nih.gov/ij/).
Total RNA was extracted and purified from Arabidopsis seeds as described (44). Real-time PCR analysis was performed using optical 96-well thin-wall plates that enabled the various Arabidopsis mutant lines to be analyzed for the presence of a specific amplicon using a single set of primers in one complete real-time PCR run. All reactions were performed in triplicate on the same plate, and the values were averaged. Reactions with the relevant cDNA negative controls and water as a template were carried out for each primer set to detect for DNA contamination. A standard curve was required for each primer set in order to carry out the data analysis. This was generated by a series of dilutions of the wild-type cDNA stock as follows: neat, 1 in 10, 1 in 100, 1 in 1000 and 1 in 10 000. In every case, 1 μL of cDNA was used as the template for the PCR reactions. As the cDNA synthesis reaction was carried out using equal quantities of RNA at the same time and using the same reagents, it has been assumed that the amount of cDNA synthesized is the same in every case as it is not possible to quantify cDNA spectrophotometrically because of the presence of the various inactivated reagents in the mix. One microlitre of cDNA was applied to the relevant wells on the 96-well plate, followed by 12.5 μL Power SYBR Green Master Mix, 1 μL of forward primer of 10 μm, 1 μL of reverse primer of 10 μm and 9.5 μL of diethyl pyrocarbonate-treated water giving a total reaction volume of 25 μL. The real-time PCR was performed using the ABI Prism 7000 Detection System. The following protocol was applied: one cycle each at 50°C for 2 min and 95°C for 10 min. Forty cycles were performed at 95°C for 15 seconds and 60°C for 1 min. Dissociation step: 95°C for 15 seconds, 60°C for 20 seconds and 95°C for 15 seconds. Product amplification efficiency of the target AtVSR isoforms was confirmed in comparison to the internal controls of tubulin in a dilution series, identical to that used for the generation of the standard curve with wild-type cDNA. Duplicate real-time PCR experiments were carried out using RNA obtained from a further extraction from the same seed stock. The data were analyzed using Abi Prism 7000 v1.2.3 software.
We are grateful to Professor Ikuko Hara-Nishimura (Kyoto University) for the kind gift of the vsr1-1 seeds. Work in the L. F. laboratory was supported in part by grant BB/C00437X from the Biotechnology and Biological Sciences Research Council (BBSRC). C. P. C. and P. R. H were funded by BBSRC (Special Studentships). Work in the D. G. R. laboratory was supported by a grant from the German Research Council (DFG Ro 440/ 13-2).